1. Introduction

Homonymy is an ordinary situation in trademark law. Two companies may exploit the same sign as long as their activities, markets, or target audiences do not overlap. Language is finite, and trademark law has long accommodated this constraint by organising the coexistence of signs through the combined operation of classes and territories.

First, the protection conferred by trademark registration is limited to the goods and services designated, in accordance with the principle of speciality. Under this principle, the Nice Classification[1] organises goods and services into 45 classes (34 for goods and 11 for services) and serves as an administrative reference tool. However, it does not, by itself, determine the material scope of protection nor the existence of a likelihood of confusion. The latter does not result solely from the identity or similarity of the signs, nor from their formal inclusion in the same class. It must be assessed globally, taking into account all relevant factors, including the similarity of the goods or services, the proximity of the relevant publics, the distinctive character of the earlier mark, and the overall impression produced by the signs.

Second, trademark protection is in principle limited to the territory for which the mark has been registered or recognised, whether national, regional (such as the European Union trademark), or international designations covering specific States. This principle of territoriality does not, however, preclude taking into account cross-border uses, including digital uses, where such uses effectively target the relevant public located within the protected territory.

Matters become more complex in the presence of domain names, since Internet naming recognises neither product classes nor sectors of activity. In addition, a given domain name, within a specific extension, can belong to only one holder. It is precisely within this space that the various extrajudicial dispute resolution mechanisms relating to domain names operate[2]. These mechanisms are designed not to reorganise existing rights as a whole, but to resolve cases of cybersquatting. Where several legitimate actors share the same sign, the central question is therefore not who is right in the abstract, but rather who is being targeted in practice.

2. Homonymy and the Internet naming system

Let us consider two homonymous trademarks, “A” and “A’”, whose coexistence raises no difficulty under trademark law. The company owning the trademark “A” manufactures bicycles. It is identified within the world of sport and soft mobility, and its mark is registered for goods in Class 12 of the Nice Classification, covering bicycles in particular. The company owning the trademark “A’”, by contrast, develops professional software and operates in the digital and immaterial services sector. Its mark is registered in distinct classes, such as Class 9 for software and Class 42 for IT services.

These two signs coexist without difficulty. They are neither competing nor substitutable, their target audiences do not overlap, and trademark law has no reason to oppose them.

The situation becomes more complex, however, when several domain names incorporating the sign “A” are registered by third parties unrelated to either company. The term is identical, but the configurations differ, and with them, public perception.

The domain name <a-software.tld> provides a first illustration. The association of the sign “A” with the term “software” immediately directs interpretation toward the digital sphere. For a reasonably attentive Internet user, this domain name suggests a publisher, a software product, or an immaterial service, most likely IT-related. Even without accessing the website, it naturally refers to the trademark “A’”, which operates in that sector, rather than to the bicycle manufacturer. Targeting here results from the very structure of the domain name.

By contrast, the domain name <a.bike> clearly falls within the universe of cycling and mobility. The .BIKE extension does not play a purely technical role. It carries an immediately intelligible sectoral meaning and may even be described as descriptive. Irrespective of any content, the domain name directs interpretation toward the economic field of the trademark “A”, registered for bicycles, and cannot reasonably be perceived as targeting the software publisher owning the trademark “A’”.

Accordingly, the owners of the trademarks “A” and “A’”, implicitly or following consultation, may each decide to initiate extrajudicial proceedings seeking the transfer or cancellation of the domain name that infringes their respective mark. Thus, the owner of trademark “A” would act against <a-software.tld>, while the owner of trademark “A’” would act against <a.bike>. In this ideal scenario, everything works smoothly.

The case of <a.online>, however, is more ambiguous. The .ONLINE extension is generic and, since digital visibility has become essential to most economic activities, it cannot be said to refer to any specific sector. As a result, the extension does not spontaneously orient interpretation toward either software or bicycles. In such a configuration, the structure of the domain name alone does not make it possible to identify the targeted trademark. The analysis must then rely on additional elements, such as the actual content of the website, its overall presentation, or the behaviour of the domain name holder. Homonymy here fully reasserts itself, and targeting cannot be presumed.

Thus, where homonymous trademarks are involved, the identity of the sign is only a starting point. What matters is not the word taken in isolation, but the way in which the domain name is constructed and how it directs public perception toward a specific economic actor.

In extrajudicial domain name proceedings, the decision-maker’s reasoning therefore does not begin with an abstract dissection of the sign. It begins with a pragmatic question: does this domain name, for a normally attentive Internet user, refer to the complainant’s trademark? Where the answer is clear, homonymy ceases to be an issue. Where it is uncertain, context becomes decisive.

3. Context and sector-specific extensions

In principle, in extrajudicial domain name proceedings, the top-level domain (.com, .online, etc.) is not taken into account when assessing the identity or similarity between a domain name and a trademark. The top-level domain (TLD) is traditionally regarded as a technical element devoid of distinctive meaning. This principle remains valid in the majority of cases and continues to serve as the starting point of the analysis. That rule, however, has limits. With the multiplication of extensions and the emergence of TLDs that the public clearly associates with particular sectors of activity[3], the role of the extension can no longer always be ignored. In certain configurations, the TLD actively contributes to the understanding of the domain name and to the identification of the targeted actor.

The example of <a.bike> illustrates this clearly. Here, the .BIKE extension does more than indicate a technical category. It explicitly refers to the world of cycling and mobility. Even in the absence of content, the domain name immediately directs interpretation toward the economic universe of trademark “A”, registered for bicycles. In such a case, the TLD directly contributes to targeting.

By contrast, <a.online> illustrates the limits of this approach. The .ONLINE extension is generic and does not refer to any particular sector. It does not, by itself, orient interpretation toward one economic universe or another. In such cases, the TLD becomes neutral again, and targeting cannot be inferred from the structure of the domain name alone. The analysis must then rely on additional elements, such as website content, the holder’s behaviour, visual elements, vocabulary used, services offered, or the manner in which the site presents itself to the public.

4. The wave of crypto-domains

It should also be emphasised that the issues analysed within the framework of traditional Internet naming (the Domain Name System, or DNS) are no longer confined to that space alone. Alternative naming systems, particularly those based on blockchain infrastructures and commonly referred to as crypto-domains or blockchain domains, are developing rapidly[4].

These systems make it possible to associate a sign with a readable identifier recorded on a public blockchain, usable as a digital identifier, a wallet address, or a gateway to decentralised services. Extensions such as .CRYPTO, .WALLET, .NFT, .DAO, or .ETH illustrate this first generation of crypto-domains, historically rooted in the crypto-asset and Web3 ecosystem.

A more recent development, however, deserves attention. Certain actors in the sector are now developing, at a rapid pace[5], extensions with a general, descriptive, or symbolic vocation, devoid of any intrinsically crypto connotation, and intended for marketing, identity, or communication purposes. New extensions such as .DREAM, .GO, .HER, .LEARN, .MOON, .PODCAST, .QUANTUM, .SECRET, .TEA, or more recently .YELLOW, are telling examples. In their economic function, these identifiers are no longer distinguishable from traditional domain names. They compete directly with the DNS as tools of visibility and targeting.

Some of these extensions, initially deployed in blockchain environments, are moreover expected to be proposed as full-fledged DNS extensions in the context of the next ICANN new gTLD application round, further blurring the boundary between traditional naming and alternative systems[6].

In these environments, homonymy is no more organised than in the DNS. Registration of a sign on a blockchain follows a logic of absolute scarcity and exclusive appropriation. Once an identifier is registered, it becomes unavailable to all other actors, including perfectly legitimate trademark owners. There is no principle of speciality, no territoriality, and no centralised regulatory mechanism comparable to those of the DNS. Existing extrajudicial mechanisms, designed to address manifest abuses in traditional naming systems, are largely ineffective or simply nonexistent[7]. The example of the trademarks “A” and “A’” could thus be replayed identically, or even more brutally, through the registration of crypto-domains, without any straightforward possibility of dispute resolution ex post.

5. Coexistence agreements as instruments of naming governance

The example of the trademarks “A” and “A’” highlights a difficulty that is often underestimated. While trademark coexistence can be legally organised, the coexistence of domain names, by contrast, is not.

Trademark law accommodates homonymy through the combined operation of goods and services classes, territories and, where appropriate, coexistence agreements. Two identical signs can therefore coexist over time, provided they do not compete and the likelihood of confusion is kept under control.

Internet naming operates according to a different logic. Within a given extension, a domain name can belong to only one holder. While joint ownership or some form of co-ownership is not, in theory, legally inconceivable, it proves ill-suited to the technical, legal and commercial functions of a domain name. As a result, co-ownership remains marginal, if not virtually nonexistent in practice.

No general rule requires, as a matter of principle, a balanced or coordinated distribution of domain names among owners of homonymous trademarks. First come, first served, regardless of any sectoral considerations. Law intervenes only ex post, where a characterised abuse is established. In practice, allocation therefore depends on largely contingent factors: the timing of registration, the level of vigilance, financial means, a more or less aggressive defensive strategy, or sheer opportunism. It is not uncommon for one actor to accumulate a substantial share of the domain names corresponding to a shared sign, without engaging in illegitimate activity or infringing anyone else’s rights. This situation may nevertheless generate lasting tension, particularly as new extensions emerge or as a sector suddenly becomes strategic.

Such a configuration can produce a structural imbalance. It does not fall within cybersquatting in the strict sense, but it shares some of its economic effects: scarcity of the resource, higher access costs to naming, and increasing dependence on defensive strategies.

The history of relations between Apple Inc. and Apple Corps Ltd. provides an emblematic illustration. For several decades, the two companies coexisted around the “Apple” marks, one in computing (Apple Inc., founded by Steve Jobs and Steve Wozniak), the other in music (Apple Corps Ltd., the Beatles’ company). This coexistence was governed by a series of settlement agreements concluded as uses evolved, until the day the two companies ended up operating in the same market. UK courts were called upon to rule on these agreements on several occasions, notably in Apple Corps Ltd v. Apple Computer Inc, where the High Court held that the marketing of music via iTunes did not, in itself, breach the parties’ contractual undertakings[8]. This decision illustrates the growing difficulty of maintaining watertight sectoral boundaries as uses converge.

When transposed to Internet naming, the issue becomes even more acute. Where trademark law can finely adjust boundaries, the domain name system imposes a binary choice: a name is taken or it is not. When one actor registers a large number of domain names corresponding to a shared sign, it does not necessarily breach trademark law, but it occupies a digital space that could, just as legitimately, have been used by another owner of the same sign.

This raises the question, in the “A” / “A’” example, whether the company holding trademark “A’” would itself be entitled to register <a.bike>. From a strictly formal standpoint, nothing would prevent it. Yet such an initiative would immediately place the domain name in a sectoral universe that does not match its actual activity. Moreover, one could imagine a sectoral expansion of the mark following the development of a hybrid invention (product and service) designed for soft mobility, in which case trademark “A’” might be extended to Class 12 (bicycles). In any event, the choice of a strongly thematic extension would orient public perception toward bicycles and mobility, that is, toward the economic universe of trademark “A”.

In such a configuration, a UDRP decision-maker[9] faced with a dispute could encounter a specific difficulty. The matter would not necessarily amount to cybersquatting in the strict sense, since the domain name holder would, by hypothesis, have rights in the sign. The dispute could instead resemble a conflict between legitimate rights holders, concerning not fraudulent appropriation, but the occupation of a digital space at the boundary of their respective spheres. Depending on the circumstances, it is not certain that the decision-maker would consider it appropriate to resolve such a dispute within an extrajudicial framework designed to sanction manifest abuses (cybersquatting). The dispute may therefore fall outside cybersquatting’s natural scope and enter a grey zone.

It is precisely in that space that coexistence agreements become concretely useful. Far from being mere tools to settle an isolated dispute, they can, when properly designed, operate as instruments of governance of the sign over time. As uses become increasingly digital and Internet naming systems become a structuring space for economic visibility, such agreements can no longer avoid addressing domain names explicitly.

A coexistence agreement does not create new rights in the DNS, but it can organise the way existing rights are exercised. It may provide priority rules, sectoral carve-outs, undertakings not to register in certain extensions, or consultation mechanisms in the event of evolving uses.

Coexistence agreements were long conceived as static instruments, meant to freeze a sectoral balance at a given point in time. Yet the digital environment is characterised by rapid changes in uses, communication channels and naming systems. In that context, an effective coexistence agreement can no longer be limited to delimiting classes of goods or territories. It must incorporate a forward-looking dimension, anticipating the emergence of new visibility supports, new extensions, and new forms of sign appropriation.

Keeping a coexistence agreement “alive” therefore requires built-in adaptation mechanisms: periodic review clauses, consultation procedures triggered by technological developments, or specific rules addressing emerging naming systems. The objective is not to rigidify uses, but to prevent contractual silence from becoming, over time, a source of conflict.

In such environments, it is hardly conceivable that owners of homonymous trademarks could simply ignore one another. Monitoring is the minimum. Knowing each other is preferable. Cooperation often becomes necessary. Far from undermining each party’s legal autonomy, limited cooperation can transform imposed homonymy into organised coexistence.

6. GlobalBlock and the effects of blocking in a context of homonymy

Tensions inherent in the coexistence of homonymous trademarks are further intensified by the emergence of global domain name blocking mechanisms, such as tools marketed under the “GlobalBlock” model. Designed to pursue a legitimate objective, namely the large-scale and automated prevention of cybersquatting, such tools profoundly alter the balance of naming where a sign is shared by several legitimate rights holders.

The logic of global blocking is straightforward. By activating such a mechanism, a trademark owner prevents third parties from registering domain names identical or very close to its sign across a very large number of extensions (often several hundred). The efficiency is real: monitoring costs decrease, the risk of cybersquatting is reduced, and protection becomes largely preventive.

In a homonymy context, however, the logic produces collateral effects. If the company holding trademark “A” activates a blocking mechanism covering the relevant extensions, the company holding trademark “A’” may find itself unable to register domain names that would otherwise be perfectly consistent with its activity, customer base and positioning. The blocking, legitimate in principle, thus becomes a source of imbalance, and potentially conflict, between rights holders whose entitlements are equivalent from a trademark-law standpoint.

The problem does not lie in manifest abuse, nor even in an intention to exclude. It lies in the nature of the tool itself. Global blocking does not distinguish between competing legitimate uses and abusive uses. It operates indiscriminately, freezing the naming space in favour of whoever moves first. In the “A” / “A’” example, the company holding trademark “A” could thus, without formally infringing any right, effectively neutralise the other company’s access to extensions that are particularly relevant to the digital sector.

This situation reveals a significant limitation of traditional coexistence agreements. For a long time, such agreements were built around sign usage, classes of goods and services, territories and, sometimes, communication media. Internet naming remained marginal, often reduced to a few “obvious” domain names. With the generalisation of global blocking mechanisms, that approach is no longer sufficient. Where a sign is shared, activating a GlobalBlock-type device can no longer realistically be treated as a purely unilateral act. It calls, de facto, for some degree of coordination. At the very least, it requires diligent upfront consideration of the blocking’s induced effects and of the restrictions it imposes on other legitimate holders of the same sign.

In a context like that of “A” and “A’”, explicitly integrating global blocking mechanisms into coexistence agreements appears to be a necessary evolution. The aim is not to prohibit such tools, nor to deny their legitimacy. It is to frame their use. The parties may, for example, agree on differentiated blocking perimeters, sectoral carve-outs, or targeted unblocking mechanisms for extensions deemed strategic for one party or the other.

Absent such contractual framing, one party’s preventive protection may, without any malicious intent, become a structural obstacle for the other. The conflict then shifts away from classic trademark infringement or cybersquatting, toward a more diffuse issue: appropriation of digital space.

Ultimately, GlobalBlock is neither a problem in itself nor a universal solution. It is a powerful tool whose use, in the presence of homonymous trademarks, must be thought through collectively. Where trademark coexistence requires legal dialogue, domain name management now requires technical and contractual dialogue. Otherwise, silence in the agreements becomes the first driver of imbalance.

7. Towards a reasoned pooling of certain resources

Finally, homonymy invites reflection on the possibility of a partial and strictly framed pooling of certain resources, particularly in the area of domain name monitoring and the fight against cybersquatting.

Detecting abusive registrations entails significant cost. It relies on tools, service providers and human resources that are often identical, even though several owners of homonymous trademarks in practice pursue a largely convergent goal: preventing opportunistic third parties from exploiting a shared sign. This duplication of efforts, rational when considered in isolation, becomes economically questionable when reproduced identically by actors who are neither competitors nor antagonists.

In a context like that of “A” and “A’”, targeted pooling of certain monitoring activities could offer clear operational value. The objective would not be to share entire protection strategies or to unify naming policies, but to focus cooperation on a narrow perimeter: detecting manifestly abusive registrations, namely those that cannot reasonably be connected to any legitimate use of the sign, whatever sector might be considered.

Such an approach, however, requires great caution. It must exclude any coordination relating to lawful uses of the sign, strategic communication choices, or the allocation of domain names among legitimate holders. Pooling can only concern the fight against clearly deviant behaviour, i.e. cybersquatting in the strict sense, and not competitive management of digital space.

The aim is therefore neither to merge rights nor to blur the legal boundaries between homonymous trademarks. It is more modest, and more realistic: to rationalise protection against a shared risk without creating strategic dependence. From that perspective, pooling appears less as a concession than as sound administration of rights, adapted to an environment where domain name scarcity and the proliferation of extensions make purely individual protection increasingly costly.

8. Conclusion

This type of issue is not unique to the DNS. Trademark law is full of examples of long-peaceful coexistences that became conflictual as uses converged and economic boundaries blurred. The opposition between Apple Inc. and Apple Corps Ltd. is an emblematic illustration. For decades, two distinct universes, technology and music, could share the same sign. But once channels, media and uses converged, homonymy alone was no longer sufficient to preserve equilibrium.

In the domain name space, this dynamic is faster and more abrupt. Extensions proliferate, some carry immediately intelligible sectoral meaning, and technical mechanisms of mass protection, such as global blocking, now make it possible to occupy or neutralise entire segments of naming. Where trademark law organises coexistence, the DNS imposes binary choices and structurally generates scarcity.

In that context, disputes involving homonymous trademarks are a reminder that extrajudicial dispute resolution mechanisms do not sanction the use of a shared sign as such. They sanction the act of using it to target, divert, or encroach upon another’s economic universe, by exploiting the domain name’s structure, its extension, its content, or the context of its registration.

Notes

[1] Nice Classification (wipo.int), established by the Nice Agreement of 15 June 1957: ompi.int.

[2] Uniform Domain Name Dispute Resolution Policy (UDRP), at the forefront : icann.org.

[3] For an overview of a sector-based classification of top-level domains, see our page “New gTLDs and .BRANDs”: https://iptwins.com/fr/nouveaux-gtld-et-marques-fr/.

[4] 2026 in Sight: CryptoTLDs, twinTLDs, and Domain Name Portfolio Strategy, iptwins.com, 2025-05-29.

[5] Ibid.

[6] Next Round Spotlight: Web2-Web3 Domain Bundles, iptwins.com, 2025-09-17; .ROBOT: A CryptoTLD Aiming for Its DNS Twin, 2025-11-04; .AGENT and Beyond: The New Frontier of Bi-Registry TLDs, iptwins.com, 20251210.

[7] Decentralized domain names: the need for discussion between operators and intellectual property right holders, iptwins.com, 2022-08-30; NFTs and intellectual property: summary of a report submitted to the U.S. Congress, iptwins.com, 2024-06-30;

[8] Apple Corps Ltd v Apple Computer Inc [2006] EWHC 996 (Ch) : bailii.org.

[9] As .BIKE is a new gTLD, disputes involving domain names under this extension fall within the scope of the UDRP.